Equine therapy is used to treat different motor disorders whose post-treatment improvements have been demonstrated, but never analyzed during horse sessions. Some processes, such as gait and parameters such as center of pressure have been analyzed qualitatively, claiming for control improvement, but never quantitatively evaluating the impact on the volunteer along different types of floors and using accelerometers. To analyze the volunteer and horse during training, we had measured their performance using three actimeters developed for the kinematic analysis of the displacement of the volunteer's head, lumbar spine and thorax; and head, neck and lumbosacral spine of the horse, through a total of six accelerometry sensors. Two trained volunteers were evaluated during a session of equine therapy walking laps on different floors: sand, grass, and concrete. The x, y and z axes of all sensors were processed in R language, to obtain the median values of each sensor according to floor, volunteer, animal, and position. It was possible to observe a significant difference between the distinct types of floors and for each position, denoting that there are differences in oscillations between the evaluation of the mounted horse, along the concrete floor with greater general variation among all axes, especially on the lumbar spine on y and x (latero-lateral and anteroposterior directions), than along the other floors. In relation to the volunteers, there was a greater lumbar oscillation y and z axes (latero-lateral and craniocaudal directions) along the concrete, and on the thorax in y and z axes along the sand floor.